Many magnetic recording systems employ a recording and playback head which has more than one transducer for a given track of the magnetic recording tape. For example, "read after write" heads have separate transducers for reading and writing data along the same track, and "read write read" heads have three transducers, a configuration which allows the head to operate as a "read after write" head in either direction of tape travel. (An erase transducer is generally included, but is not relevant here.)
Proper alignment of the transducers with respect to a reference plane of the system requires low tolerance specifications on the size and location of the transducers, as well as on the amount by which they may deviate from being parallel to each other and perpendicular to the base plane. However, even transducers manufactured within accepted tolerances may deviate from perfect alignment along data tracks. And, of course, the recording tape should also be aligned with respect to the reference plane within tolerances, but perfect alignment of the tape does not always occur. These problems are especially apparent when adding a new track, or when appending or overwriting data to an existing track, and more especially when that existing track was created by a different drive than the one in use.
The error due to improper head alignment with respect to the reference plane (known as head azimuth error) and the error due to improper tape alignment with respect to the reference plane (known as tape slope error) are in theory separate from each other, but in practice they both produce a common misalignment effect.
A full description of the geometry of head/track alignment errors and the hardware applicable to a magnetic recording system capable of compensating for such errors is found in either of U.S. Pat. Nos. 4,866,548 (Rudi) or 5,001,580 (Aranovsky et al.), each incorporated by reference into this application. The correction methods disclosed in these patents involve a special additional track that is read with the read transducer. However, these methods cannot be advantageously modified to directly apply to the write transducer, because using the write transducer in a reading mode requires modified electronic processing, at additional cost, and the wider gap width typical of write transducers may lead to poor signal quality, especially at higher data densities.